Algorithms For Stable Estimation Of The Extended State Vector Of Controlled Objects

Abstract

The article presents algorithms for stable estimation of the extended state vector of a dynamic system based on the methods of conditionally Gaussian filtering. Linearized stochastic optimal algorithms are used to solve the estimation problem. When constructing estimation algorithms, issues of stable inversion of matrices in filtration equations are considered. The above algorithms make it possible to perform a stable joint estimation of the state vector and system parameters, and thus to implement separate estimation and control subsystems.